Electromagnetic Emissions Mechanisms below 30 MHz in High Voltage Power Switching

Electromagnetic Emissions Mechanisms below 30 MHz in High Voltage Power Switching Abstract — This study investigates electromagnetic emissions mechanisms below 30 MHz due to high voltage switching of power devices. The relationship between conducted and radiated emissions is discussed. A prototype buck converter with a 300 A / 1200 V IGBT module is used to analyze switching behavior at switching speeds from 1.5 kV/μs to 28 kV/μs, using time-domain measurement methods. It is found that switching behavior affects both conducted and radiated emissions, particularly from 10 MHz to 30 MHz. Furthermore, radiation mechanisms of the buck converter are experimentally analyzed by changing its configurations. The results show that the buck converter exhibits an incidental radiated emission mechanism involving the protective earth cable, which acts as a monopole antenna due to potential fluctuations relative to the ground. The mechanism, which bypasses the conventional common mode emission path, degrades the effectiveness of choke coils and filters. Keywords — Electromagnetic emissions, IGBT, radiation mechanism, switching behavior. I. INTRODUCTION In recent years, the demand for high power semiconductor modules used in power conversion equipment, widely used in fields such as industrial applications, automotive and transportation systems, and electricity systems, has increased significantly. Insulated Gate Bipolar Transistors (IGBTs) are commonly used in these applications due to their reliability and efficiency. However, as power electronics applications expand, electromagnetic interference (EMI) has emerged as a critical concern in these fields. Various international standardization organizations have developed electromagnetic compatibility (EMC) standards to control EMI with emission limits. Such standards, including the IEC 61800-3 [1] for power drive systems, specify limits for conducted emissions in the frequency ranges from 9 kHz up to 30 MHz, and for electric radiated emissions above 30 MHz. Recently, the International Electrotechnical Commission (IEC) has initiated discussions on the standardization of requirements for magnetic radiated emissions below 30 MHz for general power electronic equipment [2]. Consequently, future EMI requirements for power electronics equipment might become stricter. To suppress the EMI, previous research has focused on filter design [3] and the parasitic capacitance of power devices [4]. Additionally, previous researchers have investigated the relationship between switching behavior and EMI using mathematical [5] and experimental methods [6], [7]. These investigations focus on conducted emissions and the spectrum of switching behavior, revealing how switching transients affect the frequency components of conducted disturbances and the optimal switching behavior for EMI mitigation. However, the relationship between switching behavior and radiated emissions below 30 MHz remains unclear, and there is limited coverage in EMC standards in this frequency range. Furthermore, the conversion mechanism from conducted to radiated emissions requires further investigation. This study analyzes EMI generation mechanisms below 30 MHz using a buck converter, which represents a fundamental topology in power electronics, as the equipment under test (EUT). The switching behavior is controlled by varying gate resistance (Rg). The relationship between conducted and radiated emissions is discussed experimentally. The radiation paths from the buck converter are investigated by changing configurations of the EUT. The rest of the paper is organized as follows: In Section II, we describe the methodology for EMI measurements, including the analysis of switching behavior. Subsequently, Section III presents the analysis results of each EMI mode at different switching speeds controlled by Rg. Then, the radiation paths from the buck converter are analyzed by changing the configuration of the EUT. Finally, Section IV summarizes how switching behavior affects conducted and radiated emissions below 30 MHz. The conversion mechanisms from conducted to radiated emissions are examined by measuring the common mode current of the input cable and protective earth (PE) cable of the EUT. The results provide insights into the radiation mechanisms of electromagnetic emissions below 30 MHz. II. METHODOLOGY The EUT is a prototype buck converter including 300 A / 1200 V IGBT module. The configuration of the EUT is shown in Fig. 1. The buck converter topology is chosen because of its simplicity, which minimizes the influence of circuit complexity on the experiments, allowing the analysis to focus directly on the EUT and the cables between the EUT and the power supply. The load is an inductor of 0.9 mH and is connected in parallel with a Free Wheeling Diode (FWD). The Yosuke Sakurai1, Jordi Solé-Lloveras23, Yasutoshi Yoshioka4, Marco A. Azpúrua23, Rik W. De Doncker5 1Device Development Department, Semiconductors Business Group, Fuji Electric Co. Ltd., Japan 2EMC Electromagnetic BCN, S.L., Spain 3Grup de Compatibilitat Electromagnètica, Universitat Politècnica de Catalunya, Spain 4Power System Control Research Department, Corporate R&D Headquarters, Fuji Electric Co. Ltd., Japan 5Institute for Power Electronics and Electrical Drives, RWTH Aachen University, Germany {sakurai-yosuke, yoshioka-yasutoshi}@fujielectric.com, {jordi.sole, marco.azpurua}@emc-barcelona.com Proc. of the 2025 International Symposium on Electromagnetic Compatibility (EMC Europe 2025), Paris, France, September 1–5, 2025 switching frequency of the IGBT, which is connected in series with the load and FWD, is fixed at 1 kHz. The duty cycle is adjusted to maintain a constant operating current during testing. The basic operating conditions include an input voltage of 600 V, a collector current (Ic) of 15 A, and a Rg of 23 Ω. To achieve faster switching, the evaluation is performed under lower current conditions than current rating of the IGBT module. Fig. 1. Buck converter for EMI measurement. The buck converter is connected with a 1.5 m input cable and a 3 m protective earth cable. 0.01–0.1 µF decoupling capacitors at the gate driver VCC–GND pins are omitted for clarity but included in the actual hardware. The parasitic capacitance between the switching-cell midpoint and PE was not directly measured in this work. A detailed experimental characterization will be carried out in future work. Measurements are performed in a 10 m semi-anechoic chamber. The measurement setup is shown in Fig. 2. A Line Impedance Stabilization Network (LISN) is used to stabilize the power supply line impedance and keep the same conditions for both conducted and radiated emission measurements. The input and PE cables of EUT are connected to the power supply through the LISN. The cables are bundled and aligned to face the antenna with the shortest distance. The setup maintains a fixed distance of 3 m between the EUT and the antenna. Fig. 2. Measurement setup for conducted and radiated emissions measurements. The experiments are performed inside a semi-anechoic chamber. This photograph shows the antenna oriented in the x direction. To analyze the switching behavior of the EUT, the Rg is adjusted to modify the switching speed. The switching behavior is shown in Fig. 3. During the turn-on event, the gate-emitter voltage (Vge) increases as the control signal for the IGBT. The collector-emitter voltage (Vce) decreases, and the collector current (Ic) starts conducting. The switching speed becomes faster with lower Rg. During the turn-on event, the upper arm FWD operates in the opposite manner as the IGBT. As the anode-cathode voltage (Vak) increases, the anode current (Ia) decreases. Comparing the switching operations of IGBT and FWD, the FWD shows a faster switching speed than the IGBT. Based on this, this study focuses on analyzing the reverse recovery behavior of FWD. The switching speed is summarized in Table . At the highest switching speed of 28 kV/μs, voltage oscillations are observed in the voltage and current waveforms of the FWD, which is directly related to an increase in electromagnetic emissions. This study examines emissions, both conducted and radiated, generated by these switching events below 30 MHz. Switching conditions of the EUT. The switching speed is controlled from average dV/dt of 28 kV/s to 1.5 kV/s by varying Rg. Average dV/dt is obtained from the slope between 10 % and 90 % of Vak. Turn-on loss (Eon) and reverse recovery loss (Err) are calculated from current and voltage waveforms given in Fig.3. Rg [Ω] Reverse recovery average dV/dt [kV/µs] Eon (Ic = 15 A) [mJ] Err (Ic = 15 A) [mJ] 0.7 28 1.83 3.31 10 6 2.78 1.98 23 4 3.48 1.58 50 1.5 4.70 1.20 For emission measurements, a time domain measurement method (shown in Fig. 4) is used instead of conventional frequency-sweep-based spectrum analysis [8], [9]. This measurement method utilizes time-domain data captured by an oscilloscope and then is processed using filters to emulate a measuring receiver. Short-Time Fourier Transform (STFT) is applied to extract frequency information. It enables frequency analysis over specific time intervals, which is useful for (a) (b) Time [µs] Time [µs] x Fig. 3. Turn on switching event (a) and reverse recovery behavior (b) of our buck converter. Switching speed is moodulated with various Rg. 1 Table 1. Proc. of the 2025 International Symposium on Electromagnetic Compatibility (EMC Europe 2025), Paris, France, September 1–5, 2025 Fig. 4. Block diagram of the multi-channel time-domain EMI measurement system used during the experiments [9]. correlating switching waveforms with EMI characteristics in this study. Additionally, the method allows to synchronize the switching events within the measurement time window. This measurement method utilizes time-domain data captured by an oscilloscope and then is processed using filters to emulate a measuring receiver. STFT is applied to extract frequency information. It enables frequency analysis over specific time intervals, which is useful for correlating switching waveforms with EMI characteristics in this study. Additionally, the method allows to synchronize the switching events within the measurement time window. III. RESULTS A. Comparison of emission modes Fig. 5 shows the emissions of the EUT under different conditions in the frequency range from 9 kHz to 30 MHz. First, conducted emissions are analyzed. These emissions are classified into two modes: differential mode and common mode. Differential mode emissions are generated by the EUT and flow through the positive and negative sides of the input cable into the LISN. Common mode emissions flow through both the input and PE cables due to parasitic capacitance between the IGBT chip and the EUT chassis. Differential mode emissions increase significantly between 5 MHz and 30 MHz as dV/dt increases from 1.6 kV/μs to 28 kV/μs. Common mode emissions remain unchanged in this frequency range. Fig. 5. Measurement results of conducted and radiated emissions as a function of switching speed (controlled by Rg). Next, the magnetic and electric radiated emissions between 5 MHz and 30 MHz show similar frequency characteristics to the differential mode emissions above 5 MHz. Generally, circuit topology should suppress radiated emissions from differential mode. However, our findings indicate that switching behavior affects both conducted and radiated emissions below 30 MHz, although traditionally only conducted emissions are considered in this frequency range. These results emphasize the importance of consideration of radiated emissions in this frequency range. B. Analysis of the radiation source of the buck converter To identify the radiation source, magnetic field radiation is measured under different configurations as shown in Fig. 6. Fig. 7 (a) shows EMI measurements with varying antenna orientations. The highest intensity is observed in the y direction between 10 MHz and 30 MHz, due to the input and PE cables radiating magnetic fields toward the antenna. Fig. 7 (b) shows results after adding a common mode choke coil to the input cable. While intensity decreases below 0.1 MHz, the y direction radiation between 10 MHz and 30 MHz shows minimal change. This result indicates that common mode current through cable loops of input and PE cable is not the main radiation source. Fig. 7 (c) shows the effect of the PE cable layout. The red line indicates results when input and PE cables are bundled together, while the blue line shows results when these cables form a horizontal loop with a 1 m diameter. Although y direction intensity remains stable despite the larger loop, x direction intensity increases unexpectedly. This indicates that radiation intensity depends more on the layout of the PE cable than on loop size. Fig. 6. Measurement configurations for analysis of the radiation sou rce. Analog Digital T1 TN … Probe or Transducer CH 1 CH N EMI Baseband digitizer ADC Pre-processing Memory Multi-domain Post-processing Spectral estimation & EMI detector emulation Multi resolution Spectrograms Waveform parameter analysis Statistical detectors … … … … VisualizationDatasets Re-analysis EMI1 EMIN Proc. of the 2025 International Symposium on Electromagnetic Compatibility (EMC Europe 2025), Paris, France, September 1–5, 2025 (a) (b) (c) (d) Fig. 7 (d) shows results without the PE cable. The radiation between 10 MHz and 30 MHz is completely suppressed. These results identify the PE cable as the primary radiation source, acting as an antenna below 30 MHz. IV. DISCUSSION Based on our analysis, switching behavior influences both conducted and radiated emissions below 30 MHz. The magnetic field is mainly emitted from the PE cable, not from the loop of the input cable and PE cable. This suggests that common mode emissions caused by the PE cable are the main contributor to the emission. However, as shown in Fig. 5, no significant components between 10 MHz and 30 MHz are observed in common mode conducted emissions. Furthermore, the common mode choke has little effect on magnetic field mitigation. To clarify this mechanism, we categorize common mode current components, as shown in Fig. 8. The line (a) in Fig. 8 shows the conventional common mode path through module parasitic capacitance and chassis, with opposite current flow in input and PE cables. The line (b) in Fig. 8 shows a unidirectional current from EUT to ground through the PE cable. Common mode currents in the input cables (positive side (INpos) and negative side (INneg)) and PE cable are measured using high-frequency current probes. The measurements are implemented under three conditions: a. The currents of INpos and INneg are measured simultaneously to observe general common mode. b. The current of PE cable is measured to observe both general common mode and common mode to ground c. INpos, INneg, and PE cable are measured simultaneously to observe common mode to ground, as general common mode is canceled in INpos and INneg. Fig. 8. Definition of common mode current between the EUT and LISN. Two common mode paths, (a) common mode (general) and (b) common mode to GND are defined depending on the current path. Fig. 9 shows the spectrum of the common mode currents in each cable at faster switching under 15 A, 0.7 Ω condition. 10 MHz component is detected in all common mode currents, particularly in the PE cable, although it is not observed in the LISN voltage measurements. This frequency component of common mode currents corresponds to the frequency component of the magnetic and electric field radiated emissions. When measuring currents of INpos, INneg, and PE cable simultaneously, the 10 MHz component is slightly reduced while the 2 MHz component is significantly reduced, indicating that the 10 MHz component is caused by common mode current to the ground. Fig. 9. Frequency components of common mode current of each cable. Fig. 7. Analysis of radiation source changing the configuration of the EUT: (a) antenna orientations; (b) installing choke coil for input cable; (c) changing PE cable layout; (d) PE cable removal. Proc. of the 2025 International Symposium on Electromagnetic Compatibility (EMC Europe 2025), Paris, France, September 1–5, 2025 These results indicate that common mode current analysis provides better insights into the relationship between conducted and radiated emissions than voltage measurements with the LISN. To analyze the common mode current to the ground in detail, the common mode current is measured both near LISN and EUT, as shown in Fig. 10. The 10 MHz component is negligible near LISN, indicating that common mode emissions to the ground are not current dependent. The current distribution pattern along the cables shows characteristics of monopole antenna standing waves, indicating that the PE cable emit radiated emission as an antenna driven by potential fluctuations of the EUT to the ground. As shown in Fig. 8, these potential fluctuations flow through the module parasitic capacitance and chassis without passing through the input cable. Reducing the parasitic capacitance between power device chip and power module package is expected to mitigate this emission path. Optimizing switching behavior also effectively reduces highfrequency components of voltage fluctuations. Fig. 10. Common mode current of each cable depending on measurement point, near EUT or near LISN. V. CONCLUSION This study investigates the relationship between conducted and radiated emissions below 30 MHz. While switching behavior traditionally affects conducted noise below 30 MHz and radiated noise above 30 MHz, our analysis reveals that switching behavior influences both conducted and radiated emissions below 30 MHz. Specifically, IGBT switching from 1.5 kV/μs to 28 kV/μs affects emissions between 10 MHz and 30 MHz. Our analysis shows that the PE cable is the primary source of radiated emissions, driven by potential fluctuations to the ground rather than typical common mode current loops. The PE cable emits radiated emission as a monopole antenna driven by potential fluctuations of the EUT to the ground. Common mode current measurements near the EUT, particularly at 10 MHz, provide better insight into this radiation mechanism than LISN voltage measurements. Conventional EMI mitigation strategies, such as choke coils or filters on input cables, are less effective for this emission pathway as they bypass the input cables. Therefore, we propose the optimization of switching transients and reduction of parasitic capacitance in semiconductor modules for electromagnetic emission reduction because the high frequency components are effectively suppressed only by optimization of switching behavior and the parasitic capacitance reduction. This study emphasizes the importance of considering both conducted and radiated emissions below 30 MHz in power electronics design, particularly for high-speed switching applications. ACKNOWLEDGMENT EMC Barcelona contribution in this paper is supported in part by the research conducted in the project “Metrology for emerging electromagnetic compatibility standards” (21NRM06 EMC-STD). This project has received funding from the European Partnership on Metrology, co-financed by the European Union’s Horizon Europe Research and Innovation Programme and by the Participating States. REFERENCES [1] International Electrotechnical Commission, "IEC 61800-3: Adjustable speed electrical power drive systems – Part 3: EMC requirements and specific test methods for power drive systems," IEC, 2022. [2] CISPR 16-2-3:2023, Specification for radio disturbance and immunity measuring apparatus and methods - Part 2-3: Methods of measurement of disturbances and immunity - Radiated disturbance measurements, International Electrotechnical Commission (IEC), Geneva, Switzerland, 2023. [3] A. Nagel et.al, “Systematic design of EMI-filters for power converters” in 2000 IEEE Industry Applications Conference, pp. 2523 – 2525, October 2000, doi: 10.1109/IAS.2000.883177. [4] K. Oberdieck et.al, “Influence of the voltage-dependent output capacitance of SiC semiconductors on the electromagnetic interference in Dc – Dc converters for electric vehicles”, in EMC EUROPE 2017, pp.1 - 6, September 2017, doi: 10.1109/EMCEurope.2017.8094652. [5] F. Costa et.al, “Graphical Analysis of the Spectra of EMI Sources in Power Electronics" IEEE TRANSACTIONS ON POWER ELECTRONICS, pp.1491 – 1498, November 2005, doi: 10.1109/TPEL.2005.857564. [6] J. Chen et.al, “Switching Transient Traced Scheme Based on Sinusoidal Waveform to Reduce EMI for IGBTs” in Proc. ECCE 2021, pp.3550 - 3554, October 2021, doi: 10.1109/ECCE47101.2021.9595912. [7] R. Zhang et.al, “Mathematical Modeling of EMI Spectrum Envelope Based on Switching Transient Behavior” IEEE Journal of Emerging and Selected Topics in Power Electronics, pp.2497 -2515, April 2022, doi: 10.1109/JESTPE.2021.3109040. [8] M. A. Azpúrua, M. Pous, M. Fernandez, and F. Silva, "Dynamic Performance Evaluation of Full Time Domain EMI Measurement Systems" in EMC EUROPE 2018, pp. 561–566, August 2018, doi: 10.1109/EMCEurope.2018.8485086. [9] J. S. Lloveras et.al, "Efficient In situ Assessment of Radiated Emissions using Time-Domain Measurements" in EMC EUROPE 2023, September 2023, doi: 10.1109/EMCEurope57790.2023.10274192. Proc. of the 2025 International Symposium on Electromagnetic Compatibility (EMC Europe 2025), Paris, France, September 1–5, 2025